INFOCOM 2013

Wakeup Scheduling for Energy-Efficient Communication in

Opportunistic Mobile Networks

Wei Gao1 and Qinghua Li2

1The University of Tennessee, Knoxville2The Pennsylvania State University

INFOCOM 2013

Outline

Introduction Big picture: energy saving through wakeup

scheduling Pairwise wakeup scheduling Cumulative wakeup scheduling Performance evaluation Conclusion

INFOCOM 2013

Opportunistic Mobile Networks

Consist of hand-held personal mobile devices Laptops, PDAs, Smartphones

Opportunistic and intermittent network connectivity Result of node mobility, device power outage, or malicious

attacks Hard to maintain end-to-end communication links

Data transmission via opportunistic contacts Communication opportunity upon physical proximity

INFOCOM 2013

Methodology of Data Transmission

Carry-and-Forward Mobile nodes physically carry data as relays Forwarding data opportunistically upon contacts Major problem: appropriate relay selection

B

A C

0.7

0.5

INFOCOM 2013

Contact Probing A node periodically probes contacts nearby

Bluetooth: discoverable mode Prerequisite for opportunistic communication

Major source of energy consumption

INFOCOM 2013

Outline

Introduction Big picture: energy saving through wakeup

scheduling Pairwise wakeup scheduling Cumulative wakeup scheduling Performance evaluation Conclusion

INFOCOM 2013

Our Focus Reduce the energy consumption of contact probing

Optimizing the probing interval is insufficient• Only avoid repetitive probing within contact duration

Energy consumption during inter-contact times• Inter-contact time is much longer than contact duration• Users only need to wake up when a contact will happen

INFOCOM 2013

Network Model Node contacts are described by the network contact

graph G(V,E) Contact process between nodes is described by

Pairwise inter-contact time is assumed to be exponentially distributed

Nodes i and j are contacted neighbors if The node set is called the contacted

neighborhood of node i

INFOCOM 2013

The Big Picture Basic idea: Wakeup scheduling

A user only wakes up when a predicted contact happens• Probabilistic contact prediction

Pairwise scheduling cumulative scheduling

INFOCOM 2013

Outline

Introduction Big picture: energy saving through wakeup

scheduling Pairwise wakeup scheduling Cumulative wakeup scheduling Performance evaluation Conclusion

INFOCOM 2013

Pairwise Wakeup Scheduling Focus: a pair of nodes A and B stay awake each time

they contact each other Tradeoff between energy saving and communication

performance A and B schedule their next wakeup period [t1, t2] at t0

Minimize the energy consumption during [t1, t2] measured

by the active ratio

Satisfy the performance requirement

Last contact between A and B

INFOCOM 2013

Minimizing the Active Ratio Contact prediction

Choose t1 and to minimize r Larger t1, larger r increases with t1

when is fixed

INFOCOM 2013

Handling Inaccuracy of Contact Prediction

Contact prediction cannot be 100% accurate A scheduled wakeup period may miss the contact Assume that the missed contact

actually happened earlier Schedule the next wakeup

period to be earlier and longer

[t0, tL] [ts, ts-dmax]

The last known contact

Maximum contact duration

INFOCOM 2013

Outline

Introduction Big picture: energy saving through wakeup

scheduling Pairwise wakeup scheduling Cumulative wakeup scheduling Performance evaluation Conclusion

INFOCOM 2013

Cumulative Wakeup Scheduling A node may have multiple contacted neighbors Cascaded integration of pairwise wakeup periods

Energy efficiency can be further optimized by integration Preserving scheduling consistency

INFOCOM 2013

Cascaded Integration Further optimization of energy efficiency

A node may contact more than one contacted neighbors during one pairwise wakeup period• E.g., A may contact C during [t1

B, t2B] ps*

• A’s pairwise wakeup period with C can be shortened due to an earlier wakeup period [t1

B, t2B]

INFOCOM 2013

Preserving Scheduling Consistency A and B may have different contacted neighborhood

Pairwise wakeup scheduling is independent Inconsistency when integrating pairwise wakeup

periods

INFOCOM 2013

Preserving Scheduling Consistency Solution: A and B communicate with each other

before cascaded integration Exchange information about pairwise wakeup periods Ensure that the pairwise wakeup period between A and B

is consistent at both A and B Relaxation is needed between A and B

Follows the one with lower contact probability

INFOCOM 2013

Outline

Introduction Big picture: energy saving through wakeup

scheduling Pairwise wakeup scheduling Cumulative wakeup scheduling Performance evaluation Conclusion

INFOCOM 2013

Simulation Setup Energy efficiency of unicast Randomization

Data generation time, TTL, data source and destinations Data forwarding strategies

Compare-and-Forward• Forward data to a relay with higher metric

Delegation• Forward data to a relay with the highest metric

Spray-and-Focus Metric for energy efficiency: avg. number of contact

probing per node during data TTL

INFOCOM 2013

Traces Realistic opportunistic mobile networks Record contacts among mobile devices when they

contact each other

INFOCOM 2013

Energy Efficiency Compare-and-forward strategy Reduces energy consumption by up to 65%

Better energy efficiency when data forwarding lasts longer Slight performance degradation

INFOCOM 2013

Impact of Cumulative Wakeup Scheduling

Cascaded integration helps further reduce energy consumption by up to another 40%

INFOCOM 2013

Conclusion Reduce energy consumption of opportunistic

communication while preserving communication performance Nodes only wake up when a predicted contact is likely to

happen Minimize the active ratio based on probabilistic

contact prediction Efficient handling of prediction inaccuracy

Integration of independent pairwise wakeup periods Further reduction of energy consumption

INFOCOM 2013

Thank you! Questions?

The paper and slides are also available at:http://web.eecs.utk.edu/~weigao